The truthiness effect depends on familiarity - the nonprobative image provided primes the brain for retrieval of facts related to the image, which then causes the individual to believe the claim because related information was easier to retrieve from memory. In exploring this phenomenon as it relates to graphics, there are (at least) four related variables which are important when considering visual displays:
Familarity
Is the provided image related to the claim, or is it unrelated?
Probative Value
Is the provided image probative; that is, does it contain the information needed to evaluate the claim?
Abstraction
The amount of information transformation necessary in order to relate the graphic to the real world. Photographs are typically low on the abstraction scale, while scatterplots and bar charts are much more abstract. Maps are an example of an intermediate abstraction image - the spatial dimensions are explicitly mapped to spatial dimensions in the real world, and any additional information must be translated visually from aesthetics such as color to numerical quantities. Abstraction is distinct from difficulty, as the same type of image, such as a map, may require different levels of mental effort in order to evaluate different claims.
This is what is manipulated in the classic truthiness paradigm: is the provided picture related or unrelated, and how does the percieved validity of the claim change based on the relatedness of the image.
In general, familiarity will be inversely related with abstraction - more abstract images will not activate stored information as easily as photographs or maps, which are less abstract.
unrelated image
This image is entirely unrelated to egypt, and thus does not provide any familiarity advantage.
pyramids
The pyramids are associated with Egypt, but provide no probative information about the shape of Egypt’s borders. Thus, this image may increase percieved familiarity without providing any probative information.
This map provides both familiarity and probative value - it is possible to examine the borders of Egypt using the map, but it also serves to remind the viewer of the shape of Egypt as well as which borders are land and coast.
This graph is probative but does not provide much additional familiarity. The increased abstraction decreases the additional information provided which may activate any related information in long term memory.
unrelated image
This image is entirely unrelated to Brazil and thus provides no familiarity activation.
Photo
The Christ the Redeemer (Christo Redentor) statue in Brazil may not be as easily associated with Brazil as the pyramids are with Egypt, so this image may not be as successful in activating familiarity.
This map of Brazil is not probative, as it provides no basis for comparison between the visual size of Brazil and the visual size of other large countries (Russia, US, Canada, China, etc.), but it activates mental information about the size of Brazil relative to South America, and the viewer may be able to use that information to recall the other countries which may be of similar size to Brazil.
Probative value is a binary variable - can the question be answered using the information provided? Classically, the truthiness paradigm only used nonprobative photos and manipulated the familiarity/relatedness of those photos to the claim.
There are different levels of effort required to pull information out of a chart or graph. For example, the mental operations necessary to evaluate the claim that “Egypt has approximately equal coastal and land borders” are as follows:
Compare this to the mental operations necessary to evaluate the same information presented in a statistical chart:
We can make this comparison require even less mental effort by collapsing the land borders into a single bar.
In the case of the Brazil example, the mental effort required to evaluate the map requires both information retrieval and potentially spatial operations. The provided map of Brazil shows the size of Brazil relative to the majority of South America. From this point, the viewer can complete either of the following sets of manipulations:
| Option 1: Primarily Information Retrieval | Option 2: Associative Retrieval |
|---|---|
| 1. “Minds eye” search for largest countries | 1. Compare Brazil to the size of South America - it takes up most of the continent |
| 2. Identify Russia, China, US, Canada, Australia as possible candidate countries | 2. Assess the size of South America relative to other continents |
| 3. Rank Brazil within the list of large nations | 3. Identify Asia, North America, Africa as larger continents than South America |
| 4. Evaluate claim | 4. Search within those continents for large countries |
| 5. Identify (Russia, China), (Canada, USA) as being large countries relative to continents that are comparably sized to South America | |
| 6. Rank candidate countries | |
| 7. Evaluate claim |
The number of steps listed here does not form a proxy for the difficulty, though, because a search of 200+ countries (or even 30+ large countries) should take longer than a search through 6 continents (Antarctica is irrelevant because it does not have any countries) - Option 2 is a recursive search strategy.
In another example, the fact “-40F is the same as -40C” can be evaluated using a thermometer with dual scales, or with a line graph.
The mental effort required to evaluate the claim using the image of the thermometer is much lower because the two scales are aligned on the same axis. The effort required to evaluate the line graph is not great, but requires locating -40 on each axis and examining whether the line passes through that point on the coordinate plane.
Quantifying the level of mental effort required to evaluate the claim using the provided information is difficult. In the case of nonprobative images, instead of quantifying the mental operations needed to transform the information provided into information which can be used to evaluate a claim, it is necessary to quantify the amount of additional information which must be retrieved in order to evaluate the claim. It may also be necessary to quantify the probability that an individual can retrieve each piece of necessary information from long term memory (if that information ever was stored in long term memory in the first place).
It may be possible to use think-aloud protocols to get at this information, and then to confirm the relative difficulty of the operation by using time-to-answer as a proxy for difficulty. In theory, more complex mental operations should take more time to execute, though this is certainly not guaranteed when evaluating cross-domain tasks; for instance, comparing retrieval from long term memory to the spatial manipulations necessary to assess the length of Egypt’s coastal border.
Abstraction: the process of considering something independently of its associations, attributes, or concrete accompaniments
In this context, images that are abstract would have lower direct semantic or “tactile” association with the topic of the claim. For instance, a map of Egypt is a much lower level of abstraction relative to either a picture of the pyramids or to a graphical chart of the numerical information.
Could an alien who didn’t know anything about our world (e.g. no associations/long term memory storage) complete the task? With higher levels of abstraction, the answer should be yes. By this definition, abstraction could be considered a measure of semantic independence.
| Fact | Ranking |
|---|---|
| Egypt has approx equal land and coastal borders | map < chart |
| Brazil is the 5th largest country | map < chart |